Copyright © 1982 Ohio Acad. Sci. 0030-0950/82/0005-0218 $2.00/0

THE POLLINATION ECOLOGY OF ORCHIS SPECTABILIS L. (ORCHIDACEAE)1

GREGG DIERINGER, Department of Biology, University of Akron, Akron, OH 44235

ABSTRACT. Three woodland sites were selected for collecting data on the frequency of pollination of Orchis spectabilis by bumblebees in northeast Ohio. A total of 82 hours was spent at 2 of these sites, which were chosen for observing the foraging o£ Bombus queens on spring ephemerals. The effect of Aesculus glabra, Dicentra canadensis, Geranium maculatum, Iris pseudacorus, and Pedicularis canadensis in attracting bumblebees in relation to 0. spectabilis pollination was also studied. Of the 123 bumblebees captured, 74% foraged on flowers and 26% hunted for nesting sites. Caging experiments on 0. spectabilis indicated 4% self-compatibility and 100% outcrossing fertility and suggested a required animal pollen vector. In the two populations studied, 17-21% of the plants flowered while fruit production ranged 0—11%. Seed production measured by two methods averaged 3485 seeds/fruit and 6656 seeds/fruit, respectively. Both methods indicated more than adequate seed set to replace the existing populations even if only one fruit/season was produced. 0. spectabilis is a woodland plant species relying on chance visits by either woodland- or meadow-dwelling Bombus queens for pollination. The limited occurrence of this plant species in northeast Ohio has not been attributed to the low pollinator frequency it receives. OHIOJ. SCI. 82(5): 218, 1982

INTRODUCTION Robertson (1924), who recorded Bombus separatus (=B. griseocollis) and B. ameri- Bombus queens emerge from hibernation = in spring and effectively pollinate a variety canorum ( B. pennsylvanicus). No mention of plant species blooming at this time of pollinial removal was made. Other ob- (Macior 1968, 1978a). Orchid flowers are servations on 0. spectabilis were made by morphologically adapted for insect pol- van Ingen (1887), Newell (1892), Ulke lination (Pijl and Dodson 1966), and (1938), and Seymour (1970) regarding slits bumblebees have been documented as pol- in the spur, pollination mechanism, and linators of some species (Ackerman 1975, color variants. Orchis can be found growing Wright 1975). Orchis spectabilis is a native throughout Ohio (Braun 1967) but seems orchid endemic to the eastern United to occur in small, disjunct populations. States. It is a member of the vernal flora Larger, more continuous populations occur growing "in rich, moist humus of beech- in protected areas of Wisconsin (Macior, maple woodlands (Auclair 1972). Gibson pers. comm.). (1897) and Shuttleworth et al. (1970) have Populations of Aesculus glabra, Dicentra speculated that 0. spectabilis is pollinated canadensis, Geranium maculatum, Iris pseuda- by bumblebees, but no field observations corus, Orchis spectabilis, and Pedicularis cana- were made. The only recorded observations densis were selected for this study. All these of 0. spectabilis visitors were made by plant species bloom in the spring and at- tract bumblebee queens as pollinators. A seasonal study of flower and bumblebee 'Manuscript received 2 June 1981 and in revised phenologies was designed to investigate form 28 September 1981 (#81-24). the pollination frequency of 0. spectabilis. 218 OhioJ. Sci. POLLINATION OF ORCHIS SPECTABILIS 219

The object of the study was to see whether were taken using artificial ultraviolet light and a 35 mm camera equipped with a fused quartz lens 0. spectabilis was receiving a low pollinator and a Chance-Pilkington OX1 ultraviolet filter. frequency relative to other spring flowers I used the Kodak Grey Scale (fig. 1) to compare in the area. If so, could this be related to the degree of reflectance. The percent reflectance its limited occurrence in northeast Ohio? equivalents are as follows: 0.00(40.5%), 0.10 Nomenclature follows Gleason and Cron- (31.5%), 0.20(19.2%), 0.30(12.0%), 0.50 (7.0%), 0.70(5.0%), 1.00(3.0%), 1.30(2.5%), quist (1963) for plants and Mitchell (1962) 1.60(2.0%), 1.90(1.5%) (Macior 1978b). for insects. Measurements were made of the tongues (prementum+glossa) of all bumblebees captured. STUDY SITES Heads of all queens captured were soaked in 10% The 3 largest populations of Orchis in potassium hydroxide overnight then rinsed with wa- ter. Tongues dissected from them were mounted in the area were chosen as sites for a compara- glycerin jelly and measured microscopically using a tive study. Site I, located near Penninsula calibrated ocular micrometer. (Summit County), contained populations The length of the corolla tube from the point of of Dicentra, Aesculus, Pedkularis, and Iris. petal fusion to the tip of the spur or tube was mea- Site II, located near Hiram (Portage sured for 6 flowers per plant species. Corbicular pollen of all bumblebees captured was County), contained mixed populations of dispersed in 70% ethanol and mounted in glycerin Geranium and Orchis. Site III, located near jelly tinted with methyl green. Pollen of the 6 plant Bath (Summit County), contained a popu- species was identified microscopically, and pollen lation of Orchis. Site I had a population of loads were classified as pure from the plant foraged Orchis too small (15 plants) for quantitative on, foreign (alien to that plant species), or mixed. Caging of Orchis spectabilis was conducted at site study. II using galvanized mesh window screen while Sites I and III were located within flowers were still in bud. A total of 14 plants with 0.5 km of farmland and open meadow, re- 46 flowers was caged; 25 flowers were artificially pollinated autogamously, one geitonogamously, and spectively. Site II bordered railroad tracks. one xenogamously. The rest were left untouched. The overstory vegetation of all sites was Extraction of pollinia was accomplished using a characteristic of a beech-maple association. pencil tip. Cages were removed at the cessation Other plant genera included Claytonia, of flowering. Dentaria, Sanguinaria, Impatiens, Trillium, The pollination mechanism of Orchis was tested Hepatica, Cimicifuga, Viburnum, Botry- 21 times using a pencil. The tip was inserted into the nectar spur and consequently came in contact chium, and Osmunda. with the rostellum breaking it open. The pollinia would become attached to the pencil via viscous pads METHODS AND MATERIALS originally located in the rostellum and could then be In the spring of 1979 and 1980, I spent a total of extracted from their enveloping sheaths. This is ex- 82 h observing insect visitors of the 6 plant species actly what happens when a bumblebee thrusts its under study at sites I and II. Of these 82 h, 34.5 h head into the nectar spur to suck nectar. A time were spent at site II. Observations in 1979 began period of 3 min was allowed for the caudicles to 3 May and ended 9 June; in 1980 they began bend the pollen sacs anteriorly and horizontally. 28 April and ended 6 June. As many bumblebees This would put the pollen sacs in a correct position were captured as possible and their behavior re- for contact with the stigmatic surface if the pencil corded. Site III provided data on the reproductive were reinserted into the flower and thereby effect biology of Orchis without capturing pollinators. pollination. Corolla reflectance for flowers of all plant species For comparative purposes, seed production per studied was determined in the laboratory using fresh fruit capsule was measured by 2 methods. For both flowers. A Bausch and Lomb Spectronic 20 reflect- methods, the capsules were split and placed in a vial ing spectrophotometer was used to measure the containing a 2 : 1 mixture of distilled water and percent reflectance of 10 selected wavelengths. A 95% ethanol and allowed to stand overnight. In Maxwell color diagram was plotted and the com- 1979 a 1—ml blow-out pipet was used to draw out posite color determined. Bumblebees possess the 1 ml of solution and place one drop on a slide. In ability to detect longwave ultraviolet light 1980 a 1—ml plastic syringe was used to draw out (Mazokhin-Porshnyakov 1962). Flowers of all spe- 0.5 ml and disperse on to quadrat ruled filter cies studied were examined for ultraviolet reflec- paper. The seeds were counted and the total seeds tance, but only those which did reflect ultraviolet produced/capsule was estimated by accounting for light (360 nm) were photographed. Photographs dilution. Only one count/capsule was made for the 220 G. DIERINGER Vol. 82

A

B FIGURE 1. (A) Ultraviolet (360 nm) reflectance of the flowero f Geranium maculatum (left), Orchis spectabilis hood and nectar spur (middle), and labellum (right). Kodak Grey Scale in background. (B) Ultraviolet (360 nm) reflectance of the fall of Iris pseudacorus. Kodak Grey Scale in background. method used in 1979 and 3 were made for the me- RESULTS thod used in 1980. Seed production was determined for capsules only at site II. FLOWER AND BUMBLEBEE PHENOLOGIES. Phenological records for plant and bumblebee The order of anthesis of the plants studied species studied were maintained. Peak blooming for for both years was as follows: Dicentra each plant species studied was only recorded in 1980 candensis, Aesculus glabra, Geranium by visual determination of the greatest number of flowers in bloom during an observation week. Week maculatum, Pedkularis canadensis, Orchis assignment corresponded to consecutive 7—day peri- spectabilis, and Iris pseudacorus. Peak ods of observation. blooming in 1980 for Dicentra was wk 2, OhioJ. Sci. POLLINATION OF ORCHIS SPECTABILIS 221 for Aeculus wk 3, for Geranium wk 5, for TABLIi 1 Pedicularis wk 4, for Orchis wk 3 and 4, Bombus queen site composition. and for /ra wk 6. All Orcvfw flowers re- mained in bloom through week 5. Total Bombus Queens Captured Only Bombus queens were captured dur- 1979 1980 ing the study; workers were not present. Site Site Site Site Queens emerged in the following order Bombus species I II I II Total % based on the earliest date captured: B. bi- affinis 3 3 4 0 10 8.1 maculatus, B. affinis, B. pagans, B. im- bimaculatus 3 0 1 1 5 4.1 patiens, B. perplexus, B. sandersoni, B. fer- fervidus 3 0 1 0 4 3.3 vidus, and B. pennsylvanicus. Bumblebee impatiens 14 12 5 1 32 26.0 species site composition is shown in table pennsylvanicus 3 0 0 0 3 2.4 perplexus 3 0 1 1 5 4.1 1. Bombus vagans comprosed 33% of the sandersoni 1 2 0 0 3 2.4 queens captured at site I in 1979 and 52% vagans 15 26 13 7 61 49.6 in 1980. At site II B. pagans comprised Total 45 43 25 10 123 100.0 60% in 1979 and 70% in 1980. The most frequent species at both sites was B. paga- ns. At site I B. pagans had a peak frequency of 3.00 bees/hr/wk during wk 6 in 1979 TABLE 2 and 2.40 bees/hr/wk during wk 5 in 1980. Total Bombus queens on plant species for sites I and II and total collecting time (hr). At site II B. pagans had a peak frequency of 1.67 bees/hr/wk during wk 3 in 1979 and 1.00 bee/hr/wk during wk 5 in 1980. Plant species Time Bombus species Total Clearly, B. pagans was the dominant bee af* bi fe im pa pe species present in the area. Aesculus Bombus ferpidus, B, pennsyhankus, and glabra 17.50 1 8 4 13 B. sandersoni were the least represented and Dicentra canadensis 21.25 4 2 1 7 were not observed hunting nesting sites. Orchis Bombus bimaculatus was not observed in this spectabilis 34.50 8 8 behavior either but Macior (1978b) reports Geranium it to be woodland-dwelling. Bees in this maculatum 34.50 1 1 2 18 22 study hunting nesting sites are considered Iris pseudacorus10.50 4 3 3 1 26 37 to be woodland-dwelling; those not ob- Pedicularis served in this behavior are considered canadensis 8.50 1 2 3 meadow- or field-dwelling. Bumblebees Total 6 5 4 12 3 4 3 53 90 will forage in both habitats no matter which is preferred for nesting. Overall, the *af — affinis, bi — bimaculatus, fe = fervidus, species composition and dominant species im = impatiens, pa = pennsylvanicus, are relatively the same so that comparisons pe = perplexus, sa = sandersoni, va = vagans between sites are meaningful. The highest frequency of B. pagans POLLINATOR ACTIVITY. Of the 123 bum- roughly corresponds to peak flowering of blebees captured (table 1), 74% were for- Geranium and Orchis at site II and of Iris at aging on flowers (table 2). Bombus pagans site I in 1980. Because of varying weather was the major pollinator or visitor to conditions from year to year, optimal polli- Geranium, Orchis, and Iris. Although not nator frequency will not correspond pre- dominant visitors to any plant studied, cisely to peak flowering of a plant species Bombus ferpidus and B. pennsyhankus to- (Schemske et al. 1978). talled 11% and 9%, respectively, of Iris Orchis had 8 Bombus pagans queens visit pollinators. its flowers; 4 of which simply landed on 222 G. DIERINGER Vol. 82 the labellum and then flew off. Only 4 TABLE 3 queens thrust their heads into the nectar Corolla tube* or spur lengths (mm). spur, and only one removed pollinia. In the other 3 instances, the pollinia had been Plant species Mean Range SD N previously removed. FLORAL ATTRACTION. From an anal- Geranium maculatum** — ysis of the visible spectral reflectance of Aesculus glabra 7 7 0.00 6 Iris pseudacorus 10 10 0.00 6 corollas of the 6 plant species, 2 major Pedicularis canadensis 11 10-12 0.75 6 divisions in visible color were noticed. Dicentra canadensis 18 16-19 0.98 6 Plant species having a peak in the violet- Orchis spectabilis 18 16-19 0.55 6 blue range (415-475 nm) included Di- centra, Geranium, and Orchis. Plant species * composed of fused or contiguous petals **no tube or spur lacking a peak in the violet-blue range and SD = standard deviation possessing one in the orange-red range N — sample size (565-655 nm) included Aesculus, Pedicu- laris, and Iris. Ultraviolet reflectance oc- curs in the corolla of Dicentra (Macior TABLE 4 1978b), the hood and nectar spur of Orchis Bombus queen tongue lengths (mm). (fig. 1A), and the fall of Iris (fig. IB). Corolla tube and spur lengths were com- Bombus species Mean Range SD N pared to Bombus queen tongue lengths (ta- bles 3 and 4). Mean values for Dicentra and sandersoni 9.40 9.21-9.78 0.33 3 Orchis show a close relationship with B. perplexus 10.35 9-00-12.69 1.67 4 affinis 10.54 8.33-12.16 1.24 9 bimaculatus and B. vagans. Both Bombus impatiens 10.59 8.33-12.80 1.43 24 species were pollinators of Dicentra but B. vagans 10.88 9.19-14.24 1.37 54 bimaculatus emerges and dominates early bimaculatus 11.28 9.33-12.69 1.45 5 during the spring. Bombus vagans emerges pennsylvanicus 13-32 11.43-14.35 1.64 3 about the same time but does not dominate fervidus 13-94 13.67-14.45 1.36 4 in frequency until later in the spring when SD = standard deviation Orchis, Geranium, and Iris are blooming. N = sample size Bombus bimaculatus was poorly represented at site II. Both Geranium and Iris attracted cave, ovate-lanceolate petals and sepals the greater percentage of B. vagans at their forming the hood. The hood is mauve in respective site. color and covers the column. Each flower is Only 7% of the Bombus queens captured subtended by a lanceolate foliar bract. carried pollen. With the exception of Iris, Three to 5 flowers per scape are born in a bees on all other plants carried mixed pol- raceme. The labellum is white, entire, and len loads. Those on Iris carried foreign pol- broadly rounded at the apex. len only. Bombus vagans was the most Sexual reproduction is accomplished by abundant pollen gatherer constituting the complex pollination mechanism pre- 56% of the total. viously described. Tests of this mechanism The inflorescence height of Geranium, using a pencil yielded 29% position Orchis, and Iris was measured for 10 plants changes capable of effecting pollination. each. Orchis inflorescence height averaged The pollinator, B. vagans, observed re- 11.8 cm and ranged 10.5-13.5 cm. Gera- moving pollinia, had 4 pollinia previously nium and Iris averaged 53.8 cm and attached medially to the frons area inferior 74.7 cm and ranged 46.4—62.9 cm and to the ocelli. Upon exiting, the bee re- 49.5-108.0 cm, respectively. moved both pollinia which became at- ORCHIS SPECTABILIS REPRODUCTIVE BI- tached to same area. The pollinia immedi- OLOGY. The flower (fig. 2) has erect, con- ately moved anteriorly and drooped with- OhioJ. Sci. POLLINATION OF ORCHIS SPECTABILIS 223

in both years. Fruit production was 9% in 1979 and 11% in 1980. Open pollination at site III was possible with 21% of the plants floweringi n 1979 and 15% in 1980. The corresponding fruit production was 1% and 0%. Of the caged flowers left untouched, none fruited, suggesting that an animal pollen vector is required. Artificial autoga- mous pollination of 25 flowersyielde d one fruit indicating a 4% self-compatibility of the flowers tested. Artificial gei- tonogamous pollination of one flower pro- duced no fruit, but artificial xenogamous pollination of one flower produced a fruit. Seed production using the 1979 method averaged 3485 seeds/fruit while the 1980 method averaged 6656 seeds/fruit. Both methods indicate more than adequate seed set to replace the existing populations even if only one fruit/season were produced. The artificially self- and cross-pollinated fruits averaged 7468 and 6954 seeds, re- spectively. These values are comparable to those produced under open pollination us- ing the 1980 method. DISCUSSION If a comparison is made of the number of FIGURE 2. Flower of Orchis spectabilis. A. Side bees captured per total observation time for view. B. Front view, a = hood, b = rostellum, each plant species studied, Orchis spectabilis c = labellum, d = ovary, e = nectar spur, received the fewest bees captured/hr while f = pollinia developing sheaths, g = stigmatic sur- face. 3*. the introduced Iris pseudacorus had the highest. It seems that Orchis may be re- out extending horizontally. Tests using a ceiving relatively few pollinators compared pencil, therefore, may give false results. to other plant species of the local flora. Other anomalies related to pollination There are 2 possible explanations for this. included the removal of only the caudicles First, ultraviolet reflectance of Iris may ef- with the pencil, the pollen sacs remaining fectively reduce pollinator frequency of Or- within the developing sheaths. Some nec- chis. The fall of Iris is highly reflective in tar spurs had ants, spiders, and slugs ultraviolet light and more importantly has blocking the entrance as well as spider a sharp contrasting pattern and large sur- webs and fungus. These obstacles might face area for attraction. Second, during effectively eliminate pollinator access. the blooming period of Orchis, which At site II there were 90 plants produc- has a relatively short inflorescence height, ing 54 flowers in 1979 and 55 in 1980. At both Geranium and Iris inflorescences are site III there were 280 plants producing at a height comparable to the surrounding 143 flowers in 1979 and 122 in 1980. vegetation. Geranium and Iris may, Open pollination of Orchis at site II was therefore, maintain a greater frequency of possible with 17% of the plants flowering bumblebees foraging consistently at this 224 G. DIERINGER Vol. 82 level. Heinrich (1975) has suggested that chance pollination. Competition from in- insects discover food sources first which troduced plants such as Iris pseudacorus and possess conspicuous flower signals. Levin pollinator activity at different vegetational and Kerster (1973) have observed that levels may be factors reducing pollination. honeybees tend to exhibit horizontal flight This low frequency pollination results in while foraging. an adequate seed production and is proba- Corbicular pollen collection, visible bly not responsible for the locally limited color reflectance, and tongue length cor- occurrence of 0. spectabilis. Further study respondence to corolla tube length are into habitat disturbance, seed viability, not directly related to maintaining an and seed germination might provide more adequate pollinator frequency. The major- useful information. ity of Bombus vagans visit flowers with ACKNOWLEDGMENTS. I would like to thank short nectar tubes. Since the spur of Orchis Dr. L. W. Macior for his help and guidance with this had nectar of various depths, spur length study. The study was partially funded by a Travail itself would not eliminate bees with Fellowship awarded by the Graduate Student Coun- short tongues. cil of the University of Akron. Orchis maintains a potential for self- pollination when a bumblebee enters, LITERATURE CITED removes pollinia, and reenters the same Ackerman, J. D. 1975 Reproductive biology of flower. A similar situation has been Goody era oblongifolia (Orchidaceae). Madrono 23: 191-198. documented with Cypripedium acaule Auclair, A. N. 1972 Comparative ecology of the (Wright 1975). The small and disjunct orchids Aplectrum byemale and Orchis spectabilis. populations combined with the capacity Bull. Torr. Bot. Club 99: 1-10. for self-pollination leaves a potential Braun, L. E. 1967 The Monocotyledoneae. Ohio for inbreeding. State Univ. Press, Columbus. Gibson, W. H. 1897 A few native orchids If pollinator frequency is the main factor and their insect sponsor. Harper's New Monthly involved in the limited occurrence of Orchis Mag. 94: 861-880. in northeast Ohio then seed set would be Gleason, H. A. and A. Cronquist 1963 Manual inadequate to maintain the populations. of vascular plants of northeastern United States This does not seem to be the case. One and adjacent Canada. D. van Nostrand Co., NY. Heinrich, B. 1975 Bee flowers: A hypothesis seed capsule/season would produce 12 X on flower variety and blooming time. Evolu- the number of seeds necessary to re- tion 29: 325-334. place the largest population studied (280 Ingen, van G. 1887 Bees mutilating flowers. plants) assuming the smaller number of Bot. Gazette 12: 229. Levin, D. A. and H. W. Kerster 1973 Assortative seeds/capsule measured. Pollinator fre- pollination for stature in Lythrum salicaria. Evo- quency, and hence seed set, seems to be lution 27: 144-152. more than adequate to maintain and even Macior, L. W. 1968 Bombus (Hymenoptera, increase the existing populations. Apidae) queen foraging in relation to vernal polli- nation in Wisconsin. Ecology 49: 20-25. Macior (unpubl. data) has recorded 1978a Pollination ecology of vernal an- Bombus fervidus, B. nevadensis auricomus, giosperms. Oikos 30: 452-460. and B. pennsylvanicus queens having Orchis 1978b Pollination interactions in sym- spectabilis pollinia attached to their heads. patric Dicentra species. Amer. J. Bot. 65: 57—62. These 3 Bombus spp. plus B. griseocollis are Mazokhin-Porshnyakov, G. A. 1962 Color- metric index of trichromic bees. Biofizika 7: considered to be open field- or meadow- 211-217. dwelling and possess tongues long enough Mitchell, T. B. 1962 Bees of the eastern United to suck nectar from Orchis (Macior 1968, States. North Car. Agric. Expt. Sta. Tech. Bull. 1978a). Bombus vagans is considered a No. 152. woodland-dwelling species. It seems that Newell, J.H. 1892 The pollination of Orchis spectabilis. Bot. Gazette 17: 165. Orchis is a woodland inhabiting plant spe- Pijl, van der L. and C. H. Dodson 1966 Orchid cies depending on both open field- and flowers: Their pollination and evolution. Univ. woodland-dwelling Bombus queens for Miami Press, FL. Ohio J. Sci. POLLINATION OF ORCHIS SPECTABILIS 225

Robertson, C. 1924 Flowers and insects. Car- Shuttleworth, F. S., H. S. Zim, G. W. Dil- linville, IL. Ion 1970 A golden guide to orchids. Western Schemske, D. W., M. F. Wilson, M. N. Melampy, Publ. Co. Inc. Golden Press, NY. L.J. Miller, L. Verner, K. M. Schemske and Ulke, T. 1938 A new white form of Orchis specta- L. B. Best 1978 Flowering ecology of some bilis L. Castanea 3: 70-71. spring woodland herbs. Ecology 59: 351—366. Wright, L. W. 1975 The pollination ecology of Seymour, F. C. 1970 Notes from the Pringle Cypripedium acaule Ait. (Orchidaceae). Unpubl. Herbarium III. Rhodora 72: 47-50. M. S. Thesis, Univ. of Akron, OH.